Received-SPF: pass (google.com: domain of linux-kernel+bounces-160835-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240427003733.3898961-1-shakeel.butt@linux.dev> <20240427003733.3898961-6-shakeel.butt@linux.dev>
In-Reply-To: <20240427003733.3898961-6-shakeel.butt@linux.dev>
From: Yosry Ahmed <yosryahmed@google.com>
Date: Fri, 26 Apr 2024 17:58:16 -0700
Message-ID: <CAJD7tkZJBBOfhHXfweJu367ov0GnppLTiUMLdoq=TcWnqu2q5w@mail.gmail.com>
Subject: Re: [PATCH v2 5/7] memcg: pr_warn_once for unexpected events and stats
To: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Andrew Morton <akpm@linux-foundation.org>, Johannes Weiner <hannes@cmpxchg.org>, 
	Michal Hocko <mhocko@kernel.org>, Roman Gushchin <roman.gushchin@linux.dev>, 
	Muchun Song <muchun.song@linux.dev>, linux-mm@kvack.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Fri, Apr 26, 2024 at 5:38=E2=80=AFPM Shakeel Butt <shakeel.butt@linux.de=
v> wrote:
>
> To reduce memory usage by the memcg events and stats, the kernel uses
> indirection table and only allocate stats and events which are being
> used by the memcg code. To make this more robust, let's add warnings
> where unexpected stats and events indexes are used.
>
> Signed-off-by: Shakeel Butt <shakeel.butt@linux.dev>
> ---
>  mm/memcontrol.c | 43 ++++++++++++++++++++++++++++++++++---------
>  1 file changed, 34 insertions(+), 9 deletions(-)
>
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 103e0e53e20a..36145089dcf5 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -671,9 +671,11 @@ unsigned long lruvec_page_state(struct lruvec *lruve=
c, enum node_stat_item idx)
>                 return node_page_state(lruvec_pgdat(lruvec), idx);
>
>         i =3D memcg_stats_index(idx);
> -       if (i >=3D 0) {
> +       if (likely(i >=3D 0)) {
>                 pn =3D container_of(lruvec, struct mem_cgroup_per_node, l=
ruvec);
>                 x =3D READ_ONCE(pn->lruvec_stats->state[i]);
> +       } else {
> +               pr_warn_once("%s: stat item index: %d\n", __func__, idx);
>         }

Can we make these more compact by using WARN_ON_ONCE() instead:

if (WARN_ON_ONCE(i < 0))
         return 0;

I guess the advantage of using pr_warn_once() is that we get to print
the exact stat index, but the stack trace from WARN_ON_ONCE() should
make it obvious in most cases AFAICT.

No strong opinions either way.